I often use calls like loglikelihood(s, arnet) where s is not strictly a Vector{Int}.
For instance, the following code fails
S = sample(arnet, 100)
julia> map(s -> loglikelihood(s, arnet), eachcol(S))
This PR makes loglikelihood a bit more versatile by changing the first argument type from Vector{Int} to AbstractVector{<:Integer}, and similarly for the function derived from ArNet
I often use calls like
loglikelihood(s, arnet)wheresis not strictly aVector{Int}. For instance, the following code failsThis PR makes
loglikelihooda bit more versatile by changing the first argument type fromVector{Int}toAbstractVector{<:Integer}, and similarly for the function derived fromArNet